识别较大图像/照片中的标记

发布于 2024-09-19 20:49:29 字数 297 浏览 3 评论 0原文

我的总体目标是在受控的类似博物馆的环境中识别相似的、反光的和低纹理的单色物体。这些物品设计精良且具有视觉吸引力。

据我了解,当我查看当代的物体和图像识别软件(例如护目镜、kooaba、tineye 等)时,这个目标太高了。

所以我正在考虑在物体上放置小标记并识别标记。 为了使用 AR 跟踪软件,标记必须是黑/白且丑陋的。这会破坏我想要识别的物体的视觉外观。

您知道有什么软件可以识别照片中的(任意设计/视觉上吸引人的)标记吗?

或者;您认为这可以通过适用于 iOS/android 设备的 openCV 来实现吗?

my overall goal is recognition of similar, reflective and low-textured monochrome objects in a controlled museum-like-environment. These objects are well designed and visually appealing.

For my understanding this goal is too high aimed when i look at contemporary object and image recognition software like goggles, kooaba, tineye etc..

So i was thinking of placing little markers on the objects and identify the markers.
In order to employ AR tracking software the markers need to be black/white and ugly. That would spoil the visual appearance of the objects i'd like to identify.

Do you know of any software that recognizes (arbitrary designed / visually appealing) markers in photographs?

Or; do you think this can be accomplished with openCV for iOS/android devices?

如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。

扫码二维码加入Web技术交流群

发布评论

需要 登录 才能够评论, 你可以免费 注册 一个本站的账号。

评论(1

桃气十足 2024-09-26 20:49:29

您已经指出了护目镜、kooaba 等的主要问题:相似性、反射、低纹理区域。

您知道有什么软件可以识别照片中的(任意设计/视觉上吸引人的)标记吗?

是的:护目镜、kooaba、tineye 等

也就是说,只要您的“任意设计/视觉上吸引人”的标记有纹理,不要太相似并且不会反射。

例如,如果您的场景是在博物馆中,玻璃后面有黑白低纹理的类似物体的图像,并且您不介意在图像旁边放置彩色标记,请尝试动物的图片(类似漫画等)。 ..通过现代图像识别设置,您可以使用这些图片作为标记。或者像信息台一样放在物体旁边,上面有一些标记。

You already name the major problems for goggles,kooaba,etc.: similarity, reflections, low-textured areas.

Do you know of any software that recognizes (arbitrary designed / visually appealing) markers in photographs?

Yes: goggles, kooaba, tineye etc

That is, as long as your "arbitrary designed / visually appealing" markers are textured, not too much similar and don´t reflect.

For example, if your scenario is in a museum with images of black/white low-textured similar objects behind glass and you don´t mind putting a colored marker next to the image, try pictures of animals (comic-like or so)... with a modern image recognition setup you can use these pictures as markers. Or put up like an information stand next to the object, which has some marker on it.

~没有更多了~
我们使用 Cookies 和其他技术来定制您的体验包括您的登录状态等。通过阅读我们的 隐私政策 了解更多相关信息。 单击 接受 或继续使用网站,即表示您同意使用 Cookies 和您的相关数据。
原文